Appendix A — Appendix: Advisory Council Deliberation
Purpose. The business logic in Business Understanding was not written by a single point of view. It is the output of a deliberate, adversarial debate among six domain specialists, each chosen to clash productively. This appendix preserves that deliberation in full so the reasoning behind the strategy — including the disagreements that were never fully resolved — is transparent and revisable.
The question on the table:
What is the best research approach for crafting a personal investing algorithm that uses geopolitical conflict and international relations as predictive signals for stocks and crypto?
The council: a Quantitative Analyst (empiricist), a Political Scientist (historian / data), an IR Theorist (strategist), a Macro / International-Business Economist (transmission-mechanism strategist), a Market-Efficiency Contrarian (devil’s advocate), and a Risk Manager (pragmatist).
A.1 Opening Positions
A.1.1 The Quantitative Analyst
Position: Build bottom-up from data — let rigorously validated patterns earn their place; narrative is where overfitting hides.
Reasoning: Markets are a feature-extraction problem. Ingest price/volume, fundamentals, and quantified event streams (GDELT, ICEWS), engineer features, and test predictive power with walk-forward, out-of-sample validation. If a “geopolitical signal” can’t survive a purged, embargoed backtest, it doesn’t exist.
Key risk they see: Everyone else will fall in love with a story about why oil spikes during invasions and never check whether the story survives the 10,000th hypothesis test.
Surprising insight: Your hardest adversary isn’t the market — it’s yourself. Most “geopolitical alpha” is data snooping dressed as insight.
A.1.2 The Political Scientist
Position: Geopolitics can be quantified — but only with the conflict-studies datasets and identification discipline the quants don’t know exist.
Reasoning: Decades of structured event data exist — GDELT, ACLED, the Correlates of War project, Polity scores, ICEWS escalation tones. These let you encode “tension,” “escalation,” and “onset” as numeric features with real construct validity. But correlation ≠ causation: you must distinguish anticipated events (priced in) from surprises (tradable).
Key risk they see: The quant will dump raw event counts into a model and learn that “more news = more volatility” — a tautology, not an edge.
Surprising insight: The tradable moment is rarely the invasion itself. It’s the slow escalation before it, when the probability shifts but consensus hasn’t caught up.
A.1.3 The IR Theorist
Position: Theory first. Without a structural model of why a conflict moves a specific market, you’re mining noise.
Reasoning: Realism predicts which conflicts threaten chokepoints (Hormuz, Taiwan Strait, Black Sea grain), which trigger alliance cascades, which stay contained. A constructivist lens tells you narrative and perception move markets as much as fundamentals. Theory points the quant’s microscope and flags spurious correlations before they cost money.
Key risk they see: A purely data-driven model will confidently “predict” the next war from patterns that were regime-specific and will never repeat. Geopolitics is brutally non-stationary.
Surprising insight: The market doesn’t react to conflict — it reacts to the revision of expectations about conflict. The real target variable is surprise, not events.
A.1.4 The Macro / International-Business Economist
Position: Trade the transmission channel, not the headline. The money is in the second-order economic consequences.
Reasoning: An invasion is an input. The tradable output is energy and grain prices, FX and safe-haven flows, defense-sector rotation, shipping/insurance costs, sanctions, and supply-chain reroutes. Russia/Ukraine wasn’t a “war trade” — it was a natural-gas, wheat, fertilizer, and defense trade with very specific tickers.
Key risk they see: Chasing the event instead of the mechanism — buying “war” instead of the three commodities and five companies actually exposed.
Surprising insight: Crypto’s role flips by regime. BTC was a “risk-on tech” asset in 2021 and a “capital-flight / sanctions” asset in 2022. The same event hits it through opposite channels depending on context — a nightmare for a static model.
A.1.5 The Market-Efficiency Contrarian
Position: Most of this edge doesn’t exist for you, and the part that does is already priced in before you can click buy.
Reasoning: Macro geopolitical information is the most-watched, best-resourced arena on earth — hedge funds with ex-intelligence analysts and sub-second feeds. By the time a conflict is legible to a personal investor, the move has happened and often mean-reverts. Survivorship bias makes the few famous “called it” trades look repeatable.
Key risk they see: The whole project becomes an elaborate machine for losing money slowly with high conviction.
Surprising insight: The one durable retail edge isn’t speed or information — it’s patience and time horizon. You can hold through volatility that forces leveraged institutions to puke. Build around that, or don’t compete at all.
A.1.6 The Risk Manager
Position: A real signal and a profitable strategy are different things. Survivability comes before returns.
Reasoning: After spread, slippage, taxes (short-term gains are brutal), and the times your “signal” is just wrong, a paper edge of a few percent evaporates. Geopolitical bets have fat tails in both directions — the conflict you predicted de-escalates overnight and your position gaps against you. Position sizing and ruin-avoidance determine the outcome more than model accuracy.
Key risk they see: One overconfident, oversized “this invasion is obvious” bet wipes out a year of disciplined gains.
Surprising insight: A mediocre signal with great risk management beats a great signal with poor risk management — every time, over a long enough horizon.
A.2 The Rebuttal Round
Quant → IR Theorist: “Your ‘theory’ is unfalsifiable. Realism, liberalism, constructivism each explain everything after the fact. I’ll take a number I can backtest over a paradigm I can’t.”
IR Theorist → Quant: “And you’ll backtest your way into bankruptcy on a correlation that held for one regime. Theory is what stops you mistaking the 2014–2022 era for a law of nature.”
Contrarian → everyone: “You’re all assuming the edge exists and arguing about how to harvest it. Has anyone proven a personal investor can extract geopolitical alpha net of costs? No? Then that’s the first study — and be prepared for the answer to be ‘no.’”
Macro Economist → Contrarian: “Half-agree. The event is priced in. The transmission is sticky — European gas dislocations and defense re-rating played out over months, not minutes. That’s a horizon a patient retail investor can actually trade.”
Political Scientist → Macro: “Only if you measure escalation before consensus. The dataset matters as much as the channel.”
Risk Manager → all: “Fascinating. Now tell me your max drawdown and what happens when you’re confidently, expensively wrong. Until then this is academic.”
A.3 Synthesis
Points of convergence (high-confidence signals)
- The headline event is the wrong target. Trade the surprise (revision of expectations) and the transmission channel — not “war.”
- Out-of-sample, cost-aware, leakage-free validation is non-negotiable. The default failure mode is fooling yourself.
- A patient, medium-term horizon is the only plausible retail edge. You cannot win the speed game against institutions.
- Risk management dominates signal quality for long-run survival.
Core tension (unresolved by design)
Top-down theory vs. bottom-up data. Theory guards against spurious, regime-specific correlations but is unfalsifiable and slow; data discovers real patterns but overfits and mistakes one era for a law. The resolution is not to pick a side but to let theory choose the features and tickers, and let data decide whether they actually predict anything — theory-guided feature engineering, validated empirically.
The blind spot (what no member fully addressed)
- Ethics — the strategy profits from invasions and human suffering; the investor should set a conscious line (see the Ethical & Personal Investment Policy section of the main chapter).
- Non-stationarity / regime change — the single biggest technical threat; the relationship found may simply stop working.
- The operator — the discipline to follow the model when it says something uncomfortable is the real bottleneck.
Recommended path
A hybrid, hypothesis-driven program: use IR/macro theory to define a small set of conflict-sensitive instruments and transmission-channel features; quantify the political layer with established event datasets; treat “geopolitical signals have tradable predictive power” as a falsifiable hypothesis the project is built to test, not assume; validate with strict walk-forward plus transaction costs; and wrap everything in a risk-first sizing framework. Build a non-geopolitical baseline so the geopolitical features must prove they add value.
Confidence level: Medium. Strong convergence on method and discipline; genuine divergence on whether the edge is large enough to matter for a personal investor — which is exactly why the project is framed to find out.
One question to sit with
If your backtest lights up with a beautiful geopolitical edge, how will you distinguish a real signal from the 10,000th lucky coin-flip — and would you stake your own capital on it before it has proven itself, untouched, on data it has never seen?
This deliberation directly shaped the confirmed project decisions recorded in Business Understanding: baseline-first framing, a Sharpe-based (risk-adjusted) success criterion, a multi-horizon blend, and an initial focus on conflict-sensitive sectors plus major crypto.